Getting Started with RepoBird

Getting Started with RepoBird

RepoBird is a powerful AI-powered GitHub automation platform that helps you solve issues, generate code, and streamline your development workflow. This guide will walk you through everything you need to know to get started.

1. Connect GitHub

The first and most crucial step is to install the RepoBird GitHub App on the repositories you want RepoBird to work with. This allows RepoBird to receive notifications about new issues and comments, and to post responses back.

Installing the GitHub App

1. Navigate to the RepoBird GitHub App installation page
2. Select the repositories you want RepoBird to access
3. Grant the necessary permissions
4. Complete the installation process

Optional: Add RepoBirdBot for Autocompletion

While not required for RepoBird to function (it uses the App installation's permissions), you can optionally add the RepoBirdBot user as a collaborator to your repositories (read access is sufficient).

Doing so enables helpful autocompletion suggestions when you start typing @RepobirdBot commands in GitHub issue or pull request comments. You can invite the bot here: https://github.com/RepoBirdBot.

2. Triggering a RepoBird Agent Run

Interacting with RepoBird involves triggering an AI agent run. You can do this by mentioning the bot handler in a comment on either a GitHub Issue or a Pull Request.

The default bot handler is @RepobirdBot.

Available Commands

Help Command

@RepobirdBot help

(or just @RepobirdBot)

Shows the help message detailing available commands and usage for the current context (Issue or Pull Request). Mentioning the bot without a command defaults to showing help.

Basic Command

@RepobirdBot basic [options] [instructions...]

Triggers a 'basic' agent run using the standard AI model.

On Issues:

• You can optionally specify source:<branch> and target:<branch>
• Any text after the command and options is added as instructions to the prompt

On Pull Requests:

• Branch options are not allowed
• You must provide instructions after the command (e.g., @RepobirdBot basic Please refactor this function.)

Pro Command

@RepobirdBot pro [options] [instructions...]

Triggers a 'pro' agent run using the advanced AI model.

On Issues:

• You can optionally specify source:<branch> and target:<branch>
• Any text after the command and options is added as instructions to the prompt

On Pull Requests:

• Branch options are not allowed
• You must provide instructions after the command (e.g., @RepobirdBot pro Implement the requested changes.)

Plan Command

@RepobirdBot plan

Generates a detailed implementation plan without writing any code. This is great for complex tasks where you want to review the approach before implementation. See our Plan Command Guide for more details.

Prompt Context

When you trigger a run from a comment, the AI agent's prompt is constructed from:

• The original Issue or Pull Request title
• The original Issue or Pull Request body
• Any additional text you provide in the triggering comment after the command itself (your specific instructions)

RepoBird will acknowledge the request in the comment thread and begin the agent run, posting status updates and eventually the results (often a new Pull Request) back to the original Issue or PR.

Pro Tips for Better Results

The AI agent does its best to research and find relevant code, files, and shell scripts by itself. However, adding context like specific file paths, code snippets, or relevant commands directly into the issue/PR body or your command comment will definitely help it run quicker and more accurately.

Be Specific About Requirements:

• Clearly mention in the issue body or command comment whether the bot should attempt to run tests or scripts (e.g., "Run npm test after changes")
• State constraints like "Do not install any new dependencies"
• Set conditions for completion, such as "Do not mark any task as done until all tests pass successfully"

This provides strong guidance for the bot's actions and leads to better results.

3. Basic vs. Pro Runs

RepoBird offers different types of processing runs, typically tied to your subscription tier. These affect the AI model used, the complexity of tasks that can be handled, and execution priority.

Pro Runs

Utilize more advanced and powerful AI models. They are better suited for complex coding tasks, detailed analysis, or generating higher-quality code. Pro runs often have higher priority in the processing queue. You might trigger these explicitly (e.g., @RepobirdBot pro) or they may be the default for higher subscription tiers.

Basic Runs

Use more cost-effective AI models, suitable for simpler tasks, quick fixes, or basic questions. These are typically available on free or lower-cost tiers and may have lower processing priority compared to Pro runs. They provide a great way to get started and handle routine automation.

Time Limits

Important: All issue runs, regardless of type, have a maximum execution time of 45 minutes.

Time Limit Behavior: If a run reaches the 45-minute limit, RepoBird will stop processing but will still attempt to create a Pull Request containing all the code changes made up to that point.

Check your subscription details or the Pricing page for specifics on run limits and features available in your tier.

Next Steps

Now that you have RepoBird set up and understand the basic commands, you're ready to start automating your development workflow! Here are some recommended next steps:

1. Start Simple: Try the plan command on a small issue to see how RepoBird analyzes your codebase
2. Review Our Guides: Check out our other documentation for advanced features and best practices
3. Join Our Community: Follow our blog for tips, updates, and success stories from other developers

If you run into any issues or have questions, don't hesitate to reach out to our support team. Happy coding!